Understanding Domestic Violence Patterns: A Problem Analysis Conducted for the Tulsa, Oklahoma Police Department
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Understanding Domestic Violence Patterns: A Problem Analysis Conducted for the Tulsa, Oklahoma Police Department December 2019 Gabrielle T. Isaza, M.S. Robin S. Engel, Ph.D. Nicholas Corsaro, Ph.D. M. Murat Ozer, Ph.D. University of Cincinnati This research was supported through a grant provided by the Laura and John Arnold Foundation (LJAF) to the International Association of Chief of Police (IACP) / University of Cincinnati (UC) Center for Police Research and Policy. The findings and recommendations presented within this report are from the authors and do not necessarily reflect the official positions or opinions of the LJAF, IACP, or Tulsa Police Department (TPD). The authors wish to thank Chief Chuck Jordan, Assistant Chief Jonathan Brooks, Charles Wulff, Tracey Lyall, and all of the officers and staff from the TPD and the Tulsa Domestic Violence Intervention Services (DVIS) for their assistance in conducting this research. Please direct all correspondence regarding this report to Dr. Robin Engel, Director, IACP/UC Center for Police Research and Policy, University of Cincinnati, 600 Teachers-Dyer Complex, 2610 McMicken Circle, Cincinnati, OH 45221-0632; 513.556.5849; robin.engel@uc.edu
Table of Contents EXECUTIVE SUMMARY ............................................................................................................ 1 I. INTRODUCTION ................................................................................................................... 5 II. ASSESSING THE EVIDENCE ON DOMESTIC VIOLENCE ......................................... 7 A. Domestic Violence in the United States ........................................................................... 7 B. Police Responses to Domestic Violence .......................................................................... 9 III. METHODOLOGY ............................................................................................................ 13 IV. PROBLEM ANALYSIS RESULTS ................................................................................. 14 A. City-Wide Domestic Violence Trends ........................................................................... 14 B. Domestic Violence Suspects .......................................................................................... 18 C. Domestic Violence Victims ........................................................................................... 22 V. DISCUSSION .................................................................................................................... 29 VI. REFERENCES .................................................................................................................. 33
EXECUTIVE SUMMARY This report summarizes the details of a problem analysis conducted by the International Association of Chiefs of Police/University of Cincinnati (IACP/UC) Center for Police Research and Policy (the “Center”) on domestic violence patterns for the Tulsa Police Department (TPD). The TPD identified domestic violence as an area requiring further understanding given the seriousness of the crime, and frequency as measured by citizen-generated calls for service. For the TPD, domestic violence-related incidents represent their third most frequent call for service, consuming substantial police resources. Prior to recommending an appropriate intervention strategy to reduce domestic violence, it is critical to better understand details regarding the specific domestic violence problem in Tulsa. For police agencies, a problem analysis includes the systematic examination of the underlying conditions of local problems they are tasked with solving. Problem analyses generally rely on information gathered from various sources, and should include examinations of both qualitative and quantitative data. The problem analysis process is critical for developing solutions that fit the problem in each community, because what works in one jurisdiction may not fit a similar problem in another setting (Boba, 2003). The following study documents a problem analysis based on a series of statistical analyses conducted on five-years of domestic violence data reported to the TPD (2013-2017) and victimization data gathered by the Family Service Center in Tulsa. Domestic violence is defined as an incidence of assault and battery against individuals connected to the suspect by one of fifteen different categories of relationship (see Oklahoma Statute §21-644). These variations in relationships cover both common (e.g. husband-wife or child-parent) and seemingly more distant (e.g. spouse-former spouse of partner or former roommate) relationships between individuals. Unfortunately, the automated data used by the TPD does not include a field for the relationship between the victim and offender. TPD data included reports for incidents, arrests, victims, calls for service, and field interview reports. These analyses were supplemented with domestic victimization data provided from the Family Service Center. The main findings of this problem analysis are summarized below. 1. Domestic violence-related calls for service were the third most frequent call received by the TPD (n=109,623; preceded by traffic stops and alarms) between 2013 and 2017. These reports were relatively stable over the five-year period, with an average of 21,942 calls per year. 2. Citizen-generated indicators of domestic violence (i.e. calls for service and incident reports) remained relatively stable over the five-year period, while the TPD-generated responses (i.e. FIR and arrest) to these crimes decreased noticeably over the same period of time. Arrests for domestic violence declined by 40.9% from 2013 to 2017, and FIRS similarly declined by 46.8% from 2013 to 2017. 1
3. TPD-generated incident reports indicate that 84.5% (n=13,381) of identified suspects were suspected of only committing one domestic violence offense, whereas 15.5% (n=2,447) were suspected of committing two or more offenses. 4. TPD-generated victimization reports indicate 83.8% (n=15,564) individuals were victimized once, and 16.2% (n=3,008) were victimized two or more times. 5. These findings demonstrate that the “repeat phenomenon” of domestic violence victimization in Tulsa is similar between victims and suspects. That is, victims are just as likely to be victimized more than once (16.2%) compared to offenders involved in more than one offense (15.5%). 6. Victimization data provided by the Family Safety Center indicates large increases in the number of services provided (180.6% increase) and the number of individuals receiving services (118.1% increase), likely due to the expansion of victim-services provided in Tulsa during the five-year study time period. Note that the increase of use of victim services is inconsistent with the downward trend in TPD domestic violence arrests. 7. Data from the FSC also indicates that the percentage of Black victims receiving FSC services (19.3%) was lower than expected given the victimization data captured by the TPD (31.6% for single incidents and 40.6% for repeat incidents), suggesting that Black domestic violence victims may be less likely to request or to receive services. 8. Analyses also show that 15.5% (n= 2,447) of the individuals suspected of domestic violence offenses were repeat offenders, and they accounted for 23.2% of the domestic violence incidents reported during the study period. 9. When offenders did commit a subsequent domestic violence offense for which an arrest occurred, the time between arrests was extremely long (an average of over 500 days). However, as the number of repeat domestic violence offenses increase per an offender, the number of days between these arrests decreases. Nonetheless, the average time between offenses is substantially long, indicating that focused deterrence approaches to repeat domestic violence offenders may not be effective in Tulsa (Sechrist, Weil, and Shelton, 2016). 10. Police data demonstrate that while a person was a victim in one incident, the same person was also commonly reported as the suspect in another domestic violence incident. Approximately 3,205 unique victims (17.2% of all domestic victims) were also charged as a suspect in a different domestic violence case. 11. Approximately 15% (8 out of 53) of offenders in domestic violence homicide incidents had a previous arrest for domestic violence. 2
12. A review of the relationship between the victim and suspect identified in the field interview reports indicates that the most common relationship type is current spouse/cohabitant (43.3%), followed by former spouse/cohabitant (20.3%), and dating (18.0%). Very few cases involve other types of relationships. 13. Field interview reports (FIR) revealed a substantial number of cases (36.7%), which involve the suspect threatening a future action—these represent an avenue for follow-up by TPD officers. The intent of this problem analysis was to provide the TPD with findings to inform the creation of a police intervention focused on domestic violence offenders. All of the findings are limited by the availability and use of official data sources. Important findings from this problem analysis—relatively low rates of repeat offenders (15%) and extremely long re-arrest periods (average of more than 500 days)—indicate that many of the well-known offender-based strategies to reduce domestic violence may not necessarily fit the domestic violence problem found in Tulsa. The results of this problem analysis demonstrate an important lesson for practitioners—it is vital to start with a problem and not with a solution. Evidence-based solutions that work elsewhere, such as the Chula Vista model (Chula Vista Police Department, 2017) and High Point model (Sechrist, Weil, and Shelton, 2016) may not fit the exact problem identified in Tulsa, Oklahoma. Interventions that focus on repeat offenders would potentially exclude the vast majority of offenders (approximately 85%) who are only arrested for domestic violence once. Despite the low levels of repeat offending, there exists a concentration of offenses amongst a smaller number of chronic offenders—these offenders represent an opportunity for future intervention by the TPD. Repeat offenders are likely to escalate in future incidents, becoming more dangerous with time (Strack, McCLane, & Hawley, 2001). Therefore, it may be more effective to focus resources on the significant few that engage higher on an offense-severity metric as opposed to an offense-frequency scale. These concerns have led the TPD to begin focusing on recording instances of non-fatal strangulation. In 2017, the TPD began training their officers to identify and record signs of victim strangulation during domestic disputes. Recent reports estimate that nearly 80% of women in Oklahoma who were victims of domestic violence between 2009 and 2013 had been strangled during that relationship (Messing et al., 2014). Indeed, non-fatal strangulation is a significant risk factor for intimate partner homicide and important for police to record when possible (Campbell, Glass, Sharps, Laughon, & Bloom, 2007). Though still in its infancy, the TPD is optimistic as to the potential benefits from its domestic strangulation initiative (Oklahoma Domestic Violence Fatality Review Board, 2018). A larger issue identified as part of this research is that much of the information used to guide domestic violence interventions is not gathered in the automated police records in Tulsa. The problem analysis provided in this report is based on the available information collected, but there 3
were several gaps in the information. For example, much of the automated data recorded by the Tulsa Police Department does not contain the victim-offender relationship. Rather, this information had to be hand-coded by Center researchers from TPD field interview reports. There is evidence to suggest that the patterns along with reasons for offending—which would impact the offender-based intervention selection—may differ according to the victim-offender relationship (see e.g., Johnson and Ferraro, 2000; Kelly and Johnson, 2008). We recommend that TPD automate information recorded on field interview reports. These reports contain valuable contextual details, such as the victim-offender relationship, suspect’s actions taken, weapons used, and threats made. The aggregation of this information would be extremely valuable to the domestic violence unit at the TPD and provide more specific details for offender- based interventions. In conclusion, the results of this problem analyses are somewhat complex. The patterns related to domestic violence offending did not match what has been seen in previous research. However, this is an important reality in policing—what works in one area may not work in another due to inherent differences in the problem. Unfortunately, domestic violence is sometimes difficult to accurately capture in police data, as offenses are often coded as an assault, without considering the victim-offender relationship. The baseline for understanding domestic violence through a policing frame is somewhat limited. There is much less research in police-led interventions for domestic violence compared to other violent crime interventions. It is evident that the City of Tulsa produces a large number of citizen-generated calls for service regarding domestic violence crimes that require police response. The most promising strategy at this point is the continuation and evaluation of the TPD strangulation initiative. Results from this exploratory analysis have provided some insight as to the specific patterns of domestic violence, but are also limited by the available data sources. 4
I. INTRODUCTION During the Spring of 2017, the International Association of Chiefs of Police/University of Cincinnati (IACP/UC) Center for Police Research and Policy entered into discussions with the Tulsa Police Department (TPD) to collaborate on a domestic violence reduction strategy. The TPD identified domestic violence as a primary area of concern because it is their third most frequent call for service, and responding to domestic violence incidents requires significant police resources. The City of Tulsa was ranked in the top quarter of U.S. Cities, with at least 250,000 residents, for violent crime (Corsaro et al., 2015). Many of these violent offenses were tied directly to domestic violence suspects. For instance, a third of the city’s homicides, half of its rapes, and nearly two-thirds of the aggravated assaults involved known domestic violence suspects (Corsaro et al., 2015). Further, the State of Oklahoma is often ranked among states with the highest rates of women killed by their domestic partners (Slipke, 2018; Violence Policy Center, 2015). Compared to other communities, the City of Tulsa has a robust and comprehensive network for providing and tracking domestic violence victim services. While there is a well-designed strategy focused on domestic violence victims, a specific strategy for police-intervention with domestic violence offenders is lacking. After a series of meetings and preliminary analyses, it was agreed that a more detailed problem analysis would benefit the TPD and allow for a better assessment of the future direction for interventions. Problem analyses are helpful for providing a comprehensive assessment of a specific crime problem to gain a more detailed understanding of the extent of the problem. The goal of problem analysis is to acquire knowledge about the characteristics and causes of police problems to understand why they occur (Boba, 2003). In order to effectively respond to a problem, police executives must first understand why a problem occurs. Using problem analysis prior to implementing evidence-based strategies is vital because it identifies the underlying patterns and factors which lead to crime and disorder problems for a police agency. Some experts have noted that while problem-oriented policing has grown in the field, problem analysis has been the slowest part of the process to develop (Boba, 2003). The Police Executive Research Forum (PERF) has found that problem analysis was the weakest phase of the problem- oriented policing process (PERF, 2000). Historically, the primary focus of police has been to respond to problems by enforcing the law or otherwise “apprehend[ing] the bad guy” (Boba, 2003). Therefore, while conducting a problem analysis is the most critical component for any crime reduction strategy, this step is most often missed. The lack of problem analyses may be due to a variety of reasons, such as reliance on anecdotes and intuition, external influences (politics, stakeholders, media, etc.), the need for expediency, or the lack of resources to conduct in-depth analyses. Law enforcement leaders must balance these constraints with the importance of using resources to fully understand the scope and context of their crime and disorder problems. Unfortunately, some law enforcement agencies fall into the trap of “finding solutions in search of problems”. This concept describes an individual or entity who may find a new and promising solution and assume it fits one or more of their problems. These new and promising evidence- 5
based strategy solutions may be discussed at gatherings of law enforcement officials, academic conferences, or read in academic or practitioner-based articles. Too often police executives or their political leadership presume that these solutions will work for their communities, without fully understanding the actual problem within their jurisdiction. It is often challenging for law enforcement leaders to convey to other stakeholders that even readily available, evidence-based solutions may not be an appropriate fit for the specific problems within their jurisdiction. In many cases, the crucial step that is missing is a problem analysis—or a systematic understanding of the problem, based on all available data. In the case of the TPD’s search for domestic violence offender-based interventions, while several evidence-based interventions appeared promising, detailed problem analyses demonstrated that these particular interventions would likely be ineffective if implemented in Tulsa. The factors associated with domestic violence in Tulsa were unique to the city relative to other settings that had implemented different strategies, and thus the foundation of the various crime prevention approaches that were conducted elsewhere were incongruent with Tulsa’s domestic violence problems. This report documents the importance of problem analysis by describing the research process conducted for the TPD examining domestic violence in their community. An enhanced description of domestic violence trends in Tulsa over the past five years was used to guide the development of a crime reduction strategy that combines specific findings from Tulsa and the most relevant evidence-based approaches from the field of criminal justice. The analyses of domestic violence patterns presented in this report are divided into three sections. Section 2 provides a brief literature review of domestic violence as a policing concern. Section 3 highlights the methodology and Section 4 includes findings from the problem analyses. These analyses demonstrate that the frequency of domestic violence related calls for service to the TPD are stable across the study time period, but field interview reports (FIR) and arrests for domestic offenses have decreased. Further, a review of the contextual details of domestic violence incidents indicates that most are characterized by low-lethality, and there are a considerably low number of repeat suspects and victims included in these reports. This report concludes in Section 5 with a discussion of the findings of the problem analysis, including suggestions for future work. 6
II. ASSESSING THE EVIDENCE ON DOMESTIC VIOLENCE Domestic violence is a specific form of violent crime where a closely defined relationship between the offender and victim is observed. Often labeled “intimate partner violence” when focused on violence between sexual partners, this concept describes physical and sexual violence, threat of physical or sexual violence, as well as psychological violence (National Institute of Justice, 2017). It should be noted that clinical definitions of domestic violence tend to be broader than legal definitions (Fantuzzo & Mohr, 1999). However, the state of Oklahoma defines domestic violence as an incidence of assault and battery against individuals connected to the suspect by one of fifteen different categories of relationship (see Oklahoma Statute §21-644). These types of relationships cover both common (e.g. husband-wife and child-parent) and seemingly more distant (e.g. spouse-former spouse of partner and former roommate) relationships between individuals. This legal definition will be used for the offenses analyzed in this report. A. Domestic Violence in the United States Historically, domestic violence was not viewed as a police concern because cultural norms suggested most events occurred in private and therefore were not something to be handled by police. In fact, throughout the 1970s, many police and legal organizations recommended that arrest be used as a last resort in dealing with family situations, with some police agencies having explicit policies against arrest in domestic disputes (Sherman, 1992). Often, this was because many incidents did not reach a level that required police intervention, or the officer was not present when the violence occurred (Sherman, 1992). Furthermore, police officers viewed domestic violence-related cases as the riskiest for officer injury, or simply as a waste of time (Buzawa, Austin & Buzawa, 1995; Buzawa & Buzawa, 1993; Ferraro, 1989). These sentiments have changed considerably over the past half century which have coincided with greater regulation of violence in the United States. A historical review of domestic violence policy provides a startling glimpse into why policy on this crime shifted. During the late 20th century: “…half of all married women will be beaten at least once by their husband [1977]... The U.S. Surgeon General found that battering of women by husbands, ex-husbands or lovers ‘is the single largest cause of injury to women in the United States’ accounting for one-fifth of all hospital emergency room cases [1989]… Thirty-one percent of all women murdered in America are killed by their husbands, ex-husbands, or lovers [1989]” (Zorza, 1992, p. 46). These findings, and the general punitive paradigm shift during the 1980s, drastically changed the United States’ approach towards domestic violence. Additionally, social movements, lawsuits against police agencies citing failure to protect (see Scott v. Hart, 1976, Bruno v. Codd, 1977, and Thurman v. Torrington, 1984), as well as research findings related to mandatory arrest (Sherman & Berk, 1984) helped give rise to this paradigm shift (Sherman, 1992). Whereas many state statutes moved to prescribe mandatory or preferred arrest of suspects of domestic disputes, Oklahoma state law currently prescribes that an officer should use their discretion in making an arrest during a domestic assault case (American Bar Association, 2014). 7
Domestic violence is not only a concern for the criminal justice system; it has been increasingly viewed as a public health crisis. Indeed, a substantial portion of research on the topic is conducted outside the discipline of criminal justice and in the domain of public health since both systems can provide first responses to these incidents. One Center for Disease Control (CDC) study estimates that in the United States, one in three women have been victims of some form of physical violence by an intimate partner within their lifetime (Breiding, Chen, & Black, 2014). Intimate partner violence is disproportionately experienced by females, racial minorities, individuals with lower incomes, and young adults (Breiding et al., 2014). A number of studies have shown that, beyond injury and death, victims of domestic violence are more likely to report a range of adverse mental and physical health conditions that may have short-term and/or long- term impact (Black, 2011; Crofford, 2007; Pico-Alfonso, Garcia-Linares, Celda-Navarro, Herbert, & Martinez, 2004). Importantly, research indicates that as the frequency of domestic violence incidents increases, the impact of the violence on the victims’ health becomes more severe (Campbell, 2002; Cox, et al., 2006). The physical and psychological toll of this crime on children in particular has several negative impacts. Research on domestic violence stresses that many of these incidents occur in homes with children (Fantuzzo & Mohr, 1999). One of the most important consequences of domestic violence is that it perpetuates the “cycle of violence”. Research indicates that child victims of domestic violence are more likely to continue perpetuating these crimes as adults (Widom, 1989). Further, children exposed to domestic violence tend to exhibit increased displays of aggressive behavior, increased emotional problems, lower levels of social competence, and poorer academic functioning (Fantuzzo & Mohr, 1999). The impact of domestic violence is far- reaching and extends beyond the immediate trauma to the victims, even influencing the economy. The financial cost of domestic violence has been estimated to exceed $8 billion per year, and in addition some estimates suggest over half the victims lose their job due to reasons associated with the incident (Rothman et al., 2007). While previous research shows mixed evidence as to whether domestic violence victims are reluctant to call the police, a study by Felson and colleagues (2002) found that three factors inhibit victims from calling the police during or following a domestic assault: desire for privacy, desire to protect the offender, and fear of reprisal from the offender. Additional research suggests that the decision-making process of reporting domestic violence for victims was more complex than previously thought, and that victims of domestic violence appear to be just as likely as other victims to report a domestic assault to police (Felson, Messner, Hoskin, & Deane, 2002). However, other researchers continue to find evidence that this crime is underreported to police (Breiding et al., 2014). For example, a more recent national study found only 36 percent of females and 12 percent of males reported their intimate partner violence victimization to police (Breiding et al., 2014). In summation, domestic violence is a difficult social problem for police to tackle due to a variety of reasons and complexities inherent in the nature of this crime. However, taking a problem- oriented approach, as proposed by Eck and Spelman (1987), necessitates an agency to critically understand any problem that they intend to tackle. The SARA process describes four stages: (1) Scanning or problem identification; (2) Analysis or investigation; (3) Response or solution 8
implementation; and (4) Assessment or evaluation (Eck & Spelman, 1987). The problem analysis used in the body of this report is an important step to devising a police-led strategy to intervene in the domestic violence problem observed in Tulsa, Oklahoma. B. Police Responses to Domestic Violence During the late 20th century a series of influential studies and court cases resulted in the adoption of mandatory and preferred arrest policies for domestic violence incidents across the country (see Sherman & Berk, 1984; Sherman 1992; Sherman & Cohn, 1989). The use of mandatory arrest in a domestic violence setting involves limiting police discretion in decision-making and directs officers to make an arrest upon arriving at a domestic violence incident, pending the specific criteria set forth in the policy. For example, many states made arrest mandatory under certain circumstances, such as victim injury or the violation of a protection order; these state policies also eliminated the requirement that an officer witness a misdemeanor offense prior to making an arrest (Brachner, 1996; Miller, 1997). Mandatory arrest policies are based on research that determined that mandatory arrest, as opposed to attempted counseling of both parties or sending the suspect away from the residence for several hours, may deter future domestic violence (Sherman & Berk, 1984). In Minneapolis, Dade County, and Colorado Springs, mandatory arrest led to fewer subsequent offenses, whereas in Omaha, Charlotte, and Milwaukee no differences in subsequent violence were observed (Garner et al., 1995). Attempts to replicate these findings in several other communities have left researchers with mixed findings about the influence of mandatory arrest on domestic violence. These replication studies also found that the deterrent effect for arrest was only significant for certain types of offenders, such as those that are employed (Sherman, 1992). Still, mandatory arrest policies were widely adopted in the 1990s (Miller, 1997). More importantly are the findings from recent research regarding mandatory arrest policies. Two studies point to significant associations between mandatory arrest and intimate partner homicides, indicating that mandatory arrest laws may have substantial unintended consequences for victims of domestic violence (Iyengar, 2009; Sherman and Harris, 2015). Police agencies explored other organizational changes to improve their response to domestic violence. A new community-oriented policing approach suggested police departments develop specialized domestic violence units or liaison officers. These units or officers work in partnership with community advocates and resources to address domestic violence in the jurisdiction using community-tailored strategies (Clarke, 1993; Sadusky, 2004). Around 76 percent of agencies reported having a program that follows up with victims after an incident using advocates, detectives, or officers, commonly referred to as “second responder” programs (Clarke, 1993; Sadusky, 2004; PERF, 2015). Research suggests these programs may increase a victim’s confidence in police, but ultimately may have little impact on reducing subsequent victimization (Davis et al., 2008; Bird, Vigurs, & Gough, 2014). Police departments can also enhance their organizational ability to identify cases and provide strong evidentiary support for the prosecution of domestic violence incidents. To aid in the identification of domestic violence cases, several agencies have turned to police training in the epidemiology of domestic violence (Chicago Police Department, 2016). While this training is 9
increasingly popular, there are no evaluations of the effectiveness of this training for identifying cases or furthering the prosecution of cases. Police agencies have also turned to enhanced technological capacity to aid in the prosecution of domestic violence cases. For example, bruise detecting LED cameras have been adopted by several agencies to capture bruising under the surface of the skin (Solis, 2014). This technology allows police to capture this information on the scene instead of having to wait for bruises to appear on a victim (Solis, 2014). Again, there is no robust empirical evidence that this tool has led to a reduction in violence or an increase in the prosecution of cases. The use of victim-based strategies provides an approach to both reducing domestic violence in cities and focusing resources to helping alleviate the traumatic impact of these crimes. The use of protection orders and emergency protection orders allows police to provide victims with immediate protection from contact with an offender (Bird, Vigurs, & Gough, 2014). One survey reports that nearly 90 percent of agencies give victims information about obtaining an order, 44 percent report helping a victim apply for an order, and nearly 30 percent report officers attending court with victims to obtain an order (PERF, 2015). Police agencies in the UK have explored 28-day protection orders to separate victims and offenders. These agencies found protection orders were associated with reductions in subsequent victimization and showed the most significant effect with repeat victims (Kelly et al., 2013); similar results have also been found in other countries (Kothari et al., 2012). Another approach to address domestic violence used in the UK and Australia involves the sharing of information about individuals who have a violent history of offending. The goal of this approach is to provide potential “high-risk” victims with information about individual offenders and services to protect themselves (Duggan, 2018). This police-led service, known generally as a “disclosure scheme,” stems from the 2009 murder of Clare Wood by her ex- partner, who had a history of domestic violence (Duggan, 2018). This scheme includes a “right to ask” component, which allows individuals to apply to the police to inquire about their partner, as well as a “right to know” aspect, which allows the proactive disclosure of this information by police or specific agencies to potential high-risk victims (Duggan, 2018; Home Office, 2016). Both disclosure approaches require case-by-case investigation to determine if disclosure is necessary, which can range from immediate disclosure to upwards of six weeks for a disclosure decision (Duggan, 2018). In the first year of the program in the UK, just over 40 percent of inquiries resulted in a disclosure (Home Office, 2016). Today there is no evidence that suggests this program is effective at reducing domestic violence but further research is underway (Fitz- Gibbon & Walklate, 2016). Both victim and offender-based domestic violence strategies have increasingly relied upon risk assessment tools to refine understanding of the individuals and address where to focus police resources. To improve the identification of individuals as “high-risk” for future domestic violence victimization, approximately 42 percent of agencies have begun using actuarial techniques (PERF, 2015). Termed “lethality assessments,” these tools consist of a short series of questions and allow officers to evaluate risk for potential victims on the scene and connect these potential victims with community resources to help mitigate risk (Campbell, Webster, & Glass, 2009). The Maryland Network Against Domestic Violence uses the Lethality Assessment Program (LAP), which involves a “lethality screen” consisting of an eleven item questionnaire to 10
determine a victim’s level of risk of death or serious injury at the hands of their partner (Lethality Assessment Program, 2019). Additionally, the LAP uses a “protocol referral” to join high-risk individuals with local domestic violence services and advocates through police (Messing et al., 2016). Although there are several tools to evaluate risk and lethality1, empirical evidence suggests these tools are effective for linking high-risk individuals with services. Subsequently, those individuals take court protective actions following the intervention both immediately and for several months after (Messing et al., 2015a; Messing, Campbell, & Wilson, 2015b). Additionally, there is evidence that tools such as the LAP program reduce the frequency and severity of domestic violence (Messing et al., 2015a). Similar to the risk/lethality assessments, one group of researchers in the UK have been exploring the use of a tool called the Priority Perpetrator Identification Tool (PPIT). The PPIT uses a short list of questions to identify individuals who may benefit from added attention from multiple agencies within a community (Economic and Social Research Council, 2017). As this tool only identifies the “priority offenders,” an intervention or program is still required to attempt to dissuade these individuals from reoffending. Recently, the use of offender-based strategies developed in the United States. These strategies considered new ways of focusing resources on offenders to reduce domestic violence. A focused deterrence model for domestic violence has stemmed from the use of focused deterrence principles to address other crime problems (Braga and Weisburd, 2012). Starting in High Point, North Carolina, this program developed a system that classifies offenders into four groups, each group receiving increasing levels of intervention based on their offending history. For example, those in the lowest group (those with no prior charges for domestic violence), are hand-delivered a letter by an officer notifying them they have been added to a “watch list” and that further violence will not be tolerated (PERF, 2015; Sumner, 2014). Those in the highest group (people with three or more prior domestic violence charges) are prosecuted immediately by any legal means possible to both provide justice for the victim, but also serve as an example to the lower group offenders about the consequences of future violence (PERF, 2015; Sumner, 2014). Evidence from this intervention suggests this method was effective in reducing the reoffending rate to 9 percent, in contrast to other studies of domestic violence, which found reoffending rates of 20 to 34 percent (Sechrist, Weil, and Shelton, 2016; PERF, 2015; Sumner, 2014). In addition, calls for service declined, but more importantly homicide by domestic violence dropped to 6 percent compared to 33 percent prior to the intervention (PERF, 2015; Sumner, 2014). Based on the successes in North Carolina, this model has been modified and expanded to several other jurisdictions across the US. For example, Lexington, North Carolina has used the exact same model as High Point, North Carolina finding a reduction in calls for service, injuries, and homicide (Sechrist, Weil, and Shelton, 2016). Additionally, Chula Vista Police Department in Chula Vista, California has modified the High Point model by adding a fifth level (for emergency intervention cases) and focusing on repeat suspects and victims (CVPD, 2017). They 1 For other tools see the Domestic Abuse, Stalking and Harassment and Honour-Based Violence Risk Identification, Assessment and Management Model (DASH) (Groves & Thomas, 2014); the Domestic Violence Risk Appraisal Guide (DVRAG) (Hilton, Harris, Rice, Houghton, & Eke, 2008); the HITS Scale (Sherin, Sinacore, Li, Zitter, & Shakil, 1998). 11
found a 24 percent reduction in subsequent offending among high-risk offenders and reduced domestic violence calls for service (CVPD, 2017). Overall, various themes have emerged from police-led interventions to domestic violence. Earlier approaches used mandatory arrest, but studies on this technique have found mixed results and even unintended negative consequences. As of 2019, several jurisdictions have moved away from the use of mandatory arrest policies. Many agencies have developed specialized units and/or liaison officers for domestic violence, but empirical assessments find this approach has little effect on reducing re-victimization. One of the more robust approaches is the use of lethality assessments to evaluate the risk of domestic violence victims—this approach has empirical support for the reduction of domestic violence victimization. And finally, another empirically supported approach is the application of focused deterrence principles to domestic violence offenders, particularly those who are chronic offenders. Multiple empirical assessments have found this approach to be effective in reducing future domestic violence offending. Despite the evidence discussed thus far, there has been limited research devoted to reducing domestic violence in policing, compared to the research devoted to other forms of violence reduction. Domestic violence is also sometimes difficult to capture accurately in police data, as offenses are often coded as an assault without considering the victim-offender relationship. In conclusion, the baseline for understanding domestic violence through a policing frame is somewhat limited. 12
III. METHODOLOGY This report includes a problem analysis of domestic violence patterns in Tulsa. The first portion of the findings section examines multiple data sources to detect trends in city-wide domestic violence patterns over time. Each of these five official data sources – calls for service, incident reports, arrests, field interview reports, and victimization data – are measured from 2013 to 2017. First, domestic violence calls for police service are generated by residents of Tulsa to notify the TPD of a possible crime occurring. Second, incident reports for domestic violence are created by TPD Officers when a citizen wants to officially record that a crime has taken place, or an officer directly observes a crime. Third, arrest reports for domestic violence incidents are made by TPD Officers when either the victim presses charges or the officers decide a suspect should be charged with a crime. Fourth, field interview reports (FIR) are a supplementary source of information generated by the TPD in addition to incident reports. A sample of approximately 422 field interview reports from 2013 were coded by IACP/UC Center for Police Research and Policy researchers to provide an overview of the contextual factors related to domestic violence incidents. These FIR were coded to compile information that is neither captured nor automated in the incident or arrest reports, including information such as victim-offender relationship, action(s) taken, suspect demeanor, etc. Finally, victimization data are provided by two sources: (1) victim reports related to domestic violence created by the Tulsa Police Department between 2013 and 2017, and (2) the Family Safety Center (FSC) in order to provide an account of domestic violence patterns through victim-generated reports that are independent from TPD and can provide additional context to the scope of the problem in Tulsa (see Davis et al., 2003; Buzawa and Buzawa, 2003). Given that domestic violence is a crime that often occurs in private spaces (i.e. households, cars, etc.), offenses are often underreported. Therefore, it is crucial for any problem analysis to consider patterns across all available data sources because assessing only one alone will not provide a comprehensive assessment of the prevalence of domestic violence in Tulsa. These data sources were provided to the IACP/UC research team as Excel workbooks and analyzed using multiple statistical software packages, including MS Excel, ArcGIS 10.3, and Stata 15.1. Descriptive statistics were used to summarize the count of each measure for the five-year observation period. The second portion of the findings section assesses characteristics of suspects and victims to describe the demographic composition of individuals involved in domestic violence. Domestic violence incident reports and victimization data were used to provide information on both individual-level patterns over time and demographic characteristics of these groups in Tulsa. Further, analyses considered and reported criminal history patterns for both suspects and victims. 13
IV. PROBLEM ANALYSIS RESULTS Due to challenges in measuring domestic violence both locally and nationally, there are only a few reliable measures to accurately provide context to Tulsa’s patterns. Using 2016 data reported from the FBI’s National Incident-Based Reporting System (NIBRS), the State of Oklahoma experienced below-average rates compared to other states in the number of violent crimes and sex offenses. However, the City of Tulsa was ranked in the top quarter of U.S. Cities with at least 250,000 residents for violent crime—many of these violent offenses were committed by known domestic violence suspects (Corsaro et al., 2015). Further, the City of Tulsa ranks in the top 10% of cities with populations over 250,000 residents in reports of rape using Uniform Crime Report (UCR) data from 2004 – 2013. Tulsa also experienced an above average number of homicides committed by individuals known to the victim, compared to these same cities according to 2014 Supplementary Homicide Report data. The State of Oklahoma is often ranked among the top worst states for rates of domestic homicide (Slipke, 2018; Violence Policy Center, 2015). Overall, despite the limitations associated with measuring domestic violence, there is some evidence that the prevalence of domestic violence in Tulsa is greater than other comparable cities, demonstrating a substantial problem with domestic violence. A. City-Wide Domestic Violence Trends Domestic violence calls for service were the first source of TPD data analyzed. Police calls for service provide both a measure of the occurrence of crime events and the reporting practices within the area. During a five-year period spanning from 2013 to 2017, 109,623 domestic violence calls for service were reported to the TPD. These calls were divided into three categories: (1) 72.7% were domestic violence incidents in progress, (2) 20.9% of incidents reportedly had just occurred, and (3) 6.4% categorized as “other”. While domestic violence calls for service accounted for only 7.2% of the total number of calls for service received by the TPD during the study period (109,623 of 1,503,326 calls for service), this frequency still represents the third most frequent call category to the TPD, with only calls for traffic stops and alarms going-off occurring more often. A low of 20,647 calls were reported in 2013 and a high of 23,437 in 2016.2 Figure 1 displays the level of consistency in the number of calls for service for each of the five years studied – a roughly 15% difference between the high and low counts is observed. 2 TPD crime analysts raised concerns about the comparability of calls for service for 2013 to 2014-2017. TPD switched CAD systems in 2013 and 2014 represents the first complete year using the current system. 14
Figure 1. Domestic Violence Calls for Service in Tulsa, 2013-2017 Domestic violence incidents reported to the TPD were the second source of TPD data analyzed. Similar to calls for service, incident reports represent both a key measure of the occurrence of crime events and the reporting practices of areas. Over the five-year study period, 23,976 incidents of domestic violence were reported to TPD. Domestic violence incidents captured only 6.9% of the total number of incidents reported for all crimes in Tulsa from 2013 – 2017 (23,976 of 349,552). Yet, domestic violence incidents still represented the 5th most frequently reported incident category to the TPD.3 Figure 2 illustrates a 10.7% increase in domestic violence incidents from 2013 to 2017. A low of 4,494 incidents was reported in 2014, and a high of 5,044 in 2017. Over the five-year study period, the total number of domestic violence incident reports represented just 22% of the total number of domestic violence calls for service4 (23,976 of 108,978 incidents). Compared to calls for service, the largest year-to-year increase occurred from 2014 to 2015 – which recorded an approximate 9.3% increase. While both measures of domestic violence recorded similar increases from 2015 to 2016, calls for service decreased from 2016 to 2017 while the number of incident reports did not. Each year the number of domestic violence incidents reported to the TPD represented approximately 20 – 25% of the total number of domestic violence calls for service. 3 The “miscellaneous” category is the most reported and traffic incidents are third. Larceny (2 nd) and burglary (4th) are the only two traditional crime categories that received more incidents reported to TPD then domestic violence. 4 Not every call for service an officer attends will result in an official police incident report; officers have some discretion in filling out a report, classifying the crime that occurred and deciding that a particular complaint is unfounded (Kruttschnitt, Kaslbeek, & House, 2014). For instance, a report from the Albuquerque Police Department demonstrated that generally, less than 20% of calls for service resulted in a report (see page 16 of http://isr.unm.edu/reports/2009/analyzing-calls-for-service-to-the-albuquerque-police-department..pdf). 15
Figure 2. Domestic Violence Incidents in Tulsa, 2013-2017 Field interview reports (FIR) and arrests were the third and fourth sources of TPD data analyzed. These measures capture TPDs response to domestic violence calls for service and possible incidents. TPD generated 17,135 field interview reports for domestic violence from 2013 to 2017. Figure 3 displays a 46.8% decrease in FIRs over this time period. There was moderate year-to-year variability in the number of FIRs with a low of 1,945 in 2017 and a high of 4,201 in 2014. Domestic violence FIRs represented 61.4% of the total FIRs documented over this study period (17,135 of 27,917 FIRs). Between 2013 – 2017, the TPD made a total of 4,811 arrests for domestic violence. Of these 4,811 arrests, 4,054 were unique individuals – 3,489 of these individuals had only one arrest (86.1% of total) while 565 had two or more (13.9% of total). 16
Figure 3. Domestic Violence FIRs in Tulsa, 2013-2017 Figure 4 illustrates a 40.9% reduction in arrests from a high of 1,170 individuals in 2013 to a low of 691 in 2017. For each of the five years studied around only 4.0% of the total number of unique individuals arrested in Tulsa were arrested for domestic violence incidents. The percent of domestic violence arrests compared to domestic violence incidents decreased from 25.7% in 2013 (1,170 arrests compared to 4,556 incidents) to 13.7% in 2017 (691 arrests compared to 5,044 incidents). 17
Figure 4. Domestic Violence Arrests in Tulsa, 2013-2017 Overall, the administrative data sources provided by the TPD begin to shape the narrative of domestic violence patterns in Tulsa from 2013 to 2017. Both domestic violence calls for service and incidents reported to TPD demonstrated consistent trends from 2013 – 2016 with a departure in 2017 when calls for service decreased and incident reports increased. Both FIRs and arrests, indicators of TPDs response to domestic violence, followed similar patterns with over a 40% reduction from 2013 to 2017 with the exception of 2013 to 2014 , when FIRs increased while arrests decreased from 2013 to 2014. Importantly, these measures of domestic violence suggest that the primarily citizen-generated indicators of domestic violence (i.e. calls for service and incident reports) remained relatively stable over the five-year period, while the TPD-generated responses (i.e. FIR and arrest) to these crimes decreased noticeably over the same period of time. B. Domestic Violence Suspects While the previous section examined city-wide trends of domestic violence, this section focuses exclusively on the individuals involved in domestic violence incidents. From 2013 – 2017, there were 18,933 individuals suspected of committing domestic violence incidents; of these, 15,828 were unique (rather than repeat) suspects. These individuals were identified from the 23,976 incident reports collected over the study period. The majority of individuals (n=13,381; 84.5% of total) were suspected of only committing one incident, whereas 2,447 (15.5%) were suspected of committing two or more incidents. Of the 2,447 individuals suspected of two or more incidents, 1,906 (12.1% of total) were involved in two incidents, 440 (2.8% of total) were involved in three incidents, and 101 (0.1% of total) were involved with more than three domestic violence offenses. These findings, displayed in Figure 5 below, suggest only a small amount of 18
domestic violence incidents are committed by repeat suspects. Repeat suspects represent 15% of the total number of suspects, which accounts for 23.2% of the domestic violence incidents. However, 84.5% of the domestic violence offenses are completed by one-time offenders, responsible for 76.8% of all domestic violence incidents between 2013 and 2017. Figure 5. Repeat Domestic Violence Offending (N=15,828) 2.8% 0.1% 12.1% One DV offense Two DV offenses Three DV offenses Four or more DV offenses 84.5% The phenomenon of repeats is well-documented in the criminological literature. Crime repeatedly occurs at the same locations, with the same victims, and is committed repeatedly by the same offenders (Spelman & Eck, 1989). In comparing what types of victimizations have the highest levels of repeats, there is evidence that family members suffer a particularly high risk of being abused again (Reiss, 1980). Repeat domestic violence offending has been demonstrated in the research (Sechrist & Weil, 2018; Williams and Houghton, 2014) and others find that domestic violence offenders have previously committed several types of offenses (Bouffard & Zedaker, 2016; Klein, 1996). Interestingly, there are fewer repeat domestic violence suspects reported to the TPD that one might expect based on other violent crime patterns (e.g., gun and gang-related violence). Examining these repeat offenders in greater detail, 1,906 committed two domestic violence offenses (77.9%), 440 individuals committed three offenses (18.0%) and 101 (4.1%) committed more than three domestic violence offenses between 2013 and 2017. The average time lapse between repeat domestic violence offenses was 582 days (median of 509 days) among those repeat domestic violence offenders that had only two domestic violence cases. The average re- offending time between the second and third domestic violence offense was 460 days (median of 407 days); between the third and fourth was 474 days (median of 414 days); and between fourth and fifth was 430 days (median of 412 days). In short, offenders with more repeat incidents have a shorter time lapse between reported incidents. Nevertheless, an average time lapse of 509 days between reported domestic violence offenses for repeat offenders is considerably longer than the 19
majority of previous research has suggested. For instance, a study of men (N=552) convicted for domestic violence, 64 percent were re-arrested for a new domestic violence offense within one- year (Klein et al., 2005). The implications of this finding for identifying an appropriate offender- based invention program will further considered in the discussion section. Below we examine different data sources, including suspect arrest data, victim information from official domestic violence incident reports, as well as homicide incident data. While each of these data sources are unique, a clear and consistent pattern of offense frequency and severity emerges across the different incidents examined below. To consider an assessment of repeat offenders, Table 1 presents the differences in demographic characteristics between suspects involved in single domestic violence incidents as well as those involved in repeat (greater than one incident over a five-year period) domestic violence incidents. There are two noteworthy demographic differences between single-incident and multiple-incident suspects. Multiple- incident suspects are both more likely to be male, and less likely to be White compared to single- incident suspects where t-tests reveal statistically significant differences (p < .01). Table 1. Demographics Characteristics of Individuals Suspected of Domestic Violence, 2013-2017 Suspects in Suspects in Single Incidents Repeat Incidents Count 13,381 2,447 Average Age 36.3 37.6 Male 72.9% 83.6% White 56.7% 43.2% Black 35.6% 51.5% Other 7.7% 5.3% When examining homicide incident data, of the 315 reported homicides during the five-year period from 2013 – 2017, 57 (18.1%) were domestic violence incidents, involving 53 unique known suspects. Of these 53 suspects identified in homicide incident data, 26.4% had a previous arrest for violence, and 49.1% had a previous arrest for a nonviolent offense in the five years prior to their arrest for homicide. Collectively, these 53 suspects averaged 1.94 total arrests in the five years prior to being arrested as a homicide suspect. Importantly, only eight of the 53 suspected individuals (15.1%) had a domestic violence arrest during the previous five years. Finally, an analysis of the field interview report sample (N=422) identified trends related to the suspects involved in domestic violence incidents. It appears that a large majority of incidents indicate low lethality—the suspect is described as calm and cooperative in 177 of the 422 incidents (41.9%), the most common description as shown in Figure 6 below. While not shown in a graphic, approximately 155 of the 422 cases (36.7%) included some type of threat by the suspect. This indicates that these cases have the potential for the suspect to return and potentially escalate the situation in a future offense. Cases where a threat is made indicate a potential area for intervention for follow up (either to the victim and/or the suspect) by the TPD. 20
You can also read